- 學位論文
基於L測度之Choquet 積分迴歸模式與赫斯特指數之耐熱蛋白預測演算法
Predicting Algorithm of Thermostable Proteins Based on Choquet Integral with Respect to L-measure and Hurst Exponent
摘要
耐熱蛋白對人類生活的影響可說是非常大。尤其在生物科技、刑事鑑定、食 品科技都是不可缺少的重要關鍵。而來尋找耐熱蛋白則是不可缺少的重要課題之 一。赫斯特指數是對分型時間序列進行敘述以及分析的方法。模糊測度和模糊積 分則是在解決多個變數之間的交互作用的方法。本篇論文是將四種非符號性的物 理化學性質數值化，並且計算赫斯特指數，進行預測分析。接著將赫斯特指數序 列進行Choquet 積分迴歸模式進行分類預測。為了要驗證這個新的演算法，我們 使用5-flod 交叉驗證法來驗證我們的演算法。實驗的結果證實我們使用的L 測 度的Choquet 積分迴歸模式比傳統的λ測度和P 測度效果來的好。同時我們也使 用複線性迴歸以及脊迴歸來比較預測效力。
並列摘要
Establishing a good algorithm for predicting temperature of thermostable proteins is an important issue. In this study, a novel thermostable proteins prediction method using Hurst exponent and Choquet integral regression model based on L-measure is proposed. The main idea of this method is to integrate the physicochemical properties, fractal property and Choquet integral regression model for amino acid symbolic sequences with different lengths. For evaluating the performance of this new algorithm, a 5-fold Cross-Validation MSE is performed. Experimental result shows that this new prediction scheme is better than the Choquet integral regression model based on λ-measure and P-measure, respectively and two methods based on Hurst exponent and the traditional prediction models, ridge regression and multiple regression model, respectively.